Think Multilingual, Not Harder: A Data-Efficient Framework for Teaching Reasoning Models to Code-Switch

• URL: http://arxiv.org/abs/2604.15490v1
• Date: Thu, 16 Apr 2026 19:44:21 GMT
• Title: Think Multilingual, Not Harder: A Data-Efficient Framework for Teaching Reasoning Models to Code-Switch
• Abstract summary: We introduce the first linguistically and behaviorally motivated fine-tuning framework for identifying beneficial code-switched reasoning behaviors in large language models.<n>First, we analyze a dataset of reasoning traces from diverse models, languages, tasks, and domains to understand the types of code-switching behaviors found in existing reasoning models.<n>We develop fine-tuning interventions that teach reasoning models to code-switch based on our observations of helpful behaviors in existing models.
• Score: 24.894866118925847
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Recent developments in reasoning capabilities have enabled large language models to solve increasingly complex mathematical, symbolic, and logical tasks. Interestingly, while reasoning models are often trained to generate monolingual text, these models have also been observed to code-switch (i.e., mix languages). Prior works have either viewed code-switching as an undesirable error, attempted to control code-switching through modifications to input prompts or the output decoding process, or focus on narrow subsets of languages, domains, tasks, and models. We address these gaps by introducing the first linguistically and behaviorally motivated fine-tuning framework for identifying beneficial code-switched reasoning behaviors in large language models and teaching these models to code-switch more effectively for reasoning. First, we create and systematically analyze a dataset of reasoning traces from diverse models, languages, tasks, and domains to understand the types of code-switching behaviors found in existing reasoning models. Then, we develop fine-tuning interventions that teach reasoning models to code-switch based on our observations of helpful behaviors in existing models. We find that our framework can significantly increase beneficial code-switched reasoning behaviors in a data-efficient manner. Interestingly, we also find that code-switching behaviors in reasoning models can be modified by fine-tuning for tasks that do not directly demonstrate code-switching in reasoning (e.g., machine translation). Our work suggests that data-efficient interventions can instill helpful forms of code-switching behavior in reasoning models.

Related papers

• The Unreasonable Effectiveness of Model Merging for Cross-Lingual Transfer in LLMs [45.08958917457921]
  Large language models (LLMs) still struggle across tasks outside of high-resource languages.<n>In this work, we investigate cross-lingual transfer to lower-resource languages where task-specific post-training data is scarce.
  arXiv  Detail & Related papers  (2025-05-23T20:28:31Z)
• Code to Think, Think to Code: A Survey on Code-Enhanced Reasoning and Reasoning-Driven Code Intelligence in LLMs [53.00384299879513]
  In large language models (LLMs), code and reasoning reinforce each other.<n>Code provides verifiable execution paths, enforces logical decomposition, and enables runtime validation.<n>We identify key challenges and propose future research directions to strengthen this synergy.
  arXiv  Detail & Related papers  (2025-02-26T18:55:42Z)
• VisualCoder: Guiding Large Language Models in Code Execution with Fine-grained Multimodal Chain-of-Thought Reasoning [10.70881967278009]
  We introduce VisualCoder, a simple yet effective approach that enhances code reasoning by integrating multimodal Chain-of-Thought snippets (CoT) reasoning with a visual Control Flow Graph (CFG)<n>We address challenges in multimodal CoT integration through a reference mechanism, ensuring consistency between code and its execution path, thereby improving performance in program behavior prediction, error detection, and output generation.
  arXiv  Detail & Related papers  (2024-10-30T19:07:01Z)
• Improving Arithmetic Reasoning Ability of Large Language Models through Relation Tuples, Verification and Dynamic Feedback [14.938401898546553]
  We propose to use a semi-structured form to represent reasoning steps of large language models. Specifically, we use relations, which are not only human but also machine-friendly and easier to verify than natural language.
  arXiv  Detail & Related papers  (2024-06-25T18:21:00Z)
• Understanding the Language Model to Solve the Symbolic Multi-Step Reasoning Problem from the Perspective of Buffer Mechanism [68.05754701230039]
  We construct a symbolic multi-step reasoning task to investigate the information propagation mechanisms in Transformer models.<n>We propose a random matrix-based algorithm to enhance the model's reasoning ability.
  arXiv  Detail & Related papers  (2024-05-24T07:41:26Z)
• Do Large Code Models Understand Programming Concepts? Counterfactual Analysis for Code Predicates [32.93686693752635]
  Large Language Models' success on text generation has also made them better at code generation and coding tasks.<n>We help bridge this gap by exploring to what degree auto-regressive models understand the logical constructs of the underlying programs.
  arXiv  Detail & Related papers  (2024-02-08T06:48:01Z)
• Code Representation Learning At Scale [75.04686476303436]
  We fuel code representation learning with a vast amount of code data via a two-stage pretraining scheme. We first train the encoders via a mix that leverages both randomness in masking language modeling and the structure aspect of programming language. We then enhance the representations via contrastive learning with hard negative and hard positive constructed in an unsupervised manner.
  arXiv  Detail & Related papers  (2024-02-02T22:19:15Z)
• Empower Nested Boolean Logic via Self-Supervised Curriculum Learning [67.46052028752327]
  We find that any pre-trained language models even including large language models only behave like a random selector in the face of multi-nested logic. To empower language models with this fundamental capability, this paper proposes a new self-supervised learning method textitCurriculum Logical Reasoning (textscClr)
  arXiv  Detail & Related papers  (2023-10-09T06:54:02Z)
• Large Language Models as Analogical Reasoners [155.9617224350088]
  Chain-of-thought (CoT) prompting for language models demonstrates impressive performance across reasoning tasks. We introduce a new prompting approach, analogical prompting, designed to automatically guide the reasoning process of large language models.
  arXiv  Detail & Related papers  (2023-10-03T00:57:26Z)
• Enhancing Zero-Shot Chain-of-Thought Reasoning in Large Language Models through Logic [19.476840373850653]
  Large language models show hallucinations as their reasoning procedures are unconstrained by logical principles. We propose LoT (Logical Thoughts), a self-improvement prompting framework that leverages principles rooted in symbolic logic. Experimental evaluations conducted on language tasks in diverse domains, including arithmetic, commonsense, symbolic, causal inference, and social problems, demonstrate the efficacy of enhanced reasoning by logic.
  arXiv  Detail & Related papers  (2023-09-23T11:21:12Z)
• Toward a Theory of Causation for Interpreting Neural Code Models [49.906221295459275]
  This paper introduces $do_code$, a post hoc interpretability method specific to Neural Code Models (NCMs) $do_code$ is based upon causal inference to enable language-oriented explanations. Results show that our studied NCMs are sensitive to changes in code syntax.
  arXiv  Detail & Related papers  (2023-02-07T22:56:58Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.